more profoundly affected by culture contact than the agricul-turalists; they typically live in biotically impoverished environ-ments, such as deserts and polar régions, as opposed to the rieh tropical and subtropical environments of many of the agricul-turalists studied; they live in smaller social groups than the agriculturalists, often numbering tens rather than thousands; and their ethnobiological knowledge has been less intensively studied than that of agriculturalists. Many of the commen-tators express suspicion of the criteria used to select cases, while several suggest that if a group of hunter-gatherers were as well studied as were many of the agriculturalists, the results would be quite comparable. These are serious criticisms, but they are largely dismissed by Brown in his reply and are not mentioned at all in this article. Since this article dépends on the earlier survey, his failure to address these problems of data quality seriously undermines its credibility.
These problems are compounded by the fact that the impor-tant variables in his model are extremely difficult to measure accurately. The démonstration that a folk category stands in a one-to-one correspondence to a scientific taxon requires com-prehensive collections of the complete biological range of each of the catégories in question. Since one discovers the range of a folk category only through careful study, casual observations (particularly of cultivated plants; see Berlin, Breedlove, and Raven 1966) will be biased toward records of one-to-one corre-spondence to scientific species. Similarly, as the earlier com-mentators Hays and Headland point out, the distinction be-tween obligatory and optional binomialism requires careful ethnolinguistic recording. Salience is the most difficult to mea-sure, as is demonstrated by the admitted crudeness of the ap-proximations that have been employed to date (e.g., Berlin et al. 1981; cf. Hays, CA 26:56-57). Yet in order to test Brown's model, one would need not only synchronie measurements of salience but also accurate measures of diachronic change in salience for entire botanical nomenclatures of a number of hunter-gatherer and agricultural societies. Given the strenuous requirements for an adequate cross-cultural sample, I believe that Brown's hypothesis is safely beyond test; studies that meet the standards of data quality required for accurate measures of even some of these variables are extremely rare and are biased toward agricultural societies. However, we do not need these data to evaluate Brown's hypothesis; it founders on internai Substantive and logical grounds.
Brown's discussion of generalization versus restriction of réf-érence in the évolution of folk generic catégories is, for the most part, a muddled restatement of Berlin (1972:58-65) with one important confusion. An ample cognitive psychological literature demonstrates that members of many natural catégo-ries are not alike in their representativeness of the category (Mervis, Catlin and Rosch 1976; Rips, Shoben, and Smith 1973; Rosch 1973, 1975; Rosch and Mervis 1975; Rosch, Simp-son, and Miller 1976). For example, though both robins and chickens are regarded as kinds of birds, robins are generally regarded as more typical than chickens. In psychological terms, the category is recognized by extension of membership or generalization from its best example, or prototype. In no-menclatural terms, however, the category label refers to the entire range of potential members. As vocabularies grow and differentiate, the internai structure of the category is explicitly recognized: extensions from the prototype are labeled with the head term plus a modifier while the prototype is often labeled simply by the head term or the head term plus the modifier "true" or "genuine" (Berlin 1972:58-65). Brown treats undif-ferentiated generic catégories as though référence of the cate-gory label were restricted to the best example, thereby con-founding psychological and nomenclatural phenomena. This confusion underlies his call to relabel terminal monolexemic catégories "spécifie" rather than "generic." In turn, this re-labeling of concepts is the most important différence between Brown and Berlin.
14
Other aspects of Brown's reformulation do not accord with common sensé. For example, hè claims that hunter-gatherers have at once smaller vocabularies and smaller referential ranges for each of their terms than agriculturalists; this means that a much larger proportion of the biological diversity that they encounter will go unnamed. Unfortunately, hè does not recognize or explain this conséquence of his position. Again, hè claims that as vocabularies expand and differentiate, the caté-gories decrease in salience; but from this it would follow that the most highly differentiated portion of folk botanical classification, cultivated plants, would be the least salient.
If errors, confusions, and relabeling of concepts are removed from Brown's reformulation, we are left with Berlin's original framework. The fact that, at base, Brown's "reformulation" is virtually indistinguishable from Berlin's framework testifies to the essential soundness of Berlin's insights.
There are problems with Berlin's framework. Some of his expectations have not been met, in particular his notion that generic catégories will not include taxa labeled by primary lexemes except for salient type-specifics. Furthermore, it is sometimes difficult to draw a clear line between taxa of differ-ent folk biological ranks, and catégories of a given rank can vary enormously in their biological range (e.g., generics range m référence from subspecies to entire botanical families). On the whole, however, both Berlin's concept of folk biological rank and his descriptions of the évolution of nomenclatural Systems have proved stimulating and useful. The exceptions to his generahzations mainly suggest the complexity of the phe-nomena ethnobiologists attempt to characterize: much of the variation m folk biological classification Systems appears to be the conséquence of the confrontation of panhuman dassifica-tory stratégies with variable biological structures
As shipwrecks mark dangerous shoals, the problems in Brown s attempted reformulation should instruct us. Some are inherent m his methods of cross-cultural comparison; others appear to stem from a failure to think through the implications of his position. Until the procedures of ethnobiological research become more standardized and the data base greatly ex-panded, I believe we can learn more from carefully controlled compansons.of Systems than from large uncritical surveys. Progress will depend on more careful measurement and analy-sis of the great variety of potential explanatory variables men-tioned above.
by THILO C. SCHADEBERG
Department of African Linguistics, University of Leiden, Stationsplein 10, P.O. Box 9515, 2300 RA Leiden, The Netherlands. 8 vu 85
The kind of research reported in Brown's article is of great potential importance for the method of reconstructing semantic change. Histoncal hnguistics has reached a high degree of sophistication and formalization where the reconstruction of forms such as sounds, words, and sentences is concerned This contrasts with the more intuitive appeal to plausibility that is commonly invoked when a particular (change of) meaning is reconstructed (see Bynon 1977:62). It appears to be common opinion that replacement, extension, and restriction of a word's possible set of referents are all possible changes (see Jeffers and Lehiste 1979:127-29). The introduction of genera! pnnciples that would narrow the range of possible semantic changes—at least for ethnobiological nomenclature—would greatly increase the accuracy of semantic reconstruction. We might then hope to overcome the major stumbling block to the application of the "Wörter und Sachen" method for hy-pothesizing the homeland of speakers of a reconstructed pro-tolanguage.
ways primary and that this kind of "polysemy" invariably re-sul ts from extending the set of referents. A documented counterexample (see the Oxford English Dictionary) is the his-tory of the English word deer: "tl. A beast; usually a quad-ruped—1481. 2. The genera! name of a family (Cervidae)." The intermediate stage of "polysemy" is attested by Shake-speare. "Mice, Rats, and such small Deare "
There is also interesting évidence from first-language acqui-sition where the generic use of a word predates its restriction to more spécifie uses. A German baby, for instance, may pass through a stage in which its whole universe is classifled into
Mama, Wauwau, and Auto. I know of a small boy who for a
time called a lamp suspended from the ceiling Engel 'angel'. This happened shortly after Christmas, when the family had had a Christmas tree decorated with a tinsel angel, and for a time his vocabulary contained an item Engel 'suspended, dangling object'.
I believe that this situation is typical for human language. The common type of "polysemy" in which a given word refers typically to its most typical referents but also—if the need arises—to similar referents for which no conventional more generic label exists may well be the primary situation par ex-cellence. Unless we push speakers of other languages to use équivalents of our own populär or scientific classes of referents (e.g., "banana," "kangaroo/wallaby"), we shall probably find it very difficult to distinguish the first two stages that Brown postulâtes for the émergence of generic/type-specific polysemy (in both patterns).
More specifically, I would like to challenge Brown's state-ment "Wood/tree polysemy is invariably the result of extending a term for a high-salience referent, wood, to a much lower-salience referent, tree." In Bantu, where the same "polysemy" is attested, the meaning "tree" is almost certainly the more basic one. The item "tree" is formally marked as belonging to a grammatical gender which has sometimes been called "tree gender" because it contains not only the word for "tree" but also many spécifie tree names. Some Bantu languages hâve dissolved thé "original polysemy" by using thé same root as for "tree" but assigning it to a différent gender when referring to "wood," e.g., Kinga umu-biki 'tree', iki-biki 'wood', i.e., "something of the nature of trees."
With regard to Brown's statement "Overt marking usually, if
not always, develops from polysemy," I would like to point out
that there are many "binomial labels" that must have been created without any intermediate "polysemy" stage (see, for example, the 20-odd terms sea X, where X = anémone,
but-terfly, cow, etc.).
Brown's observation that small-scale agriculturalists have larger vocabularies for wild plants and animais than hunter-gatherers surprises me. Peoples depending on cattle have rieh related vocabularies; peoples that keep bees have large apicul-tural terminologies. I had always assumed that hunter-gatherers possessed unexcelled knowledge about their ecolog-ical habitats and that their languages encoded this knowledge. As far as Africa is concerned, I doubt whether this point can profitably be investigated by counting terms in existing dic-tionaries. Linguistic knowledge about agriculturalists is gener-ally much more extensive than about hunter-gatherers.
by LEONTINE E. VISSER
Institute of Cultural and Social Studies, University of Leiden, Stationsplein W, P.O. Box 9507, 2300 RA Leiden, The Netherlands. 3 Vil 85
Brown's article follows the lines of thought set out by Berlin but also refines them. Apparently thé mode of subsistence, whether it is hunting-gathering or small-scale agriculture, is crucial for ethnobiological classification. Binomialization oc-curs far more frequently among agriculturalists and, according to Brown, results from the diminishing salience of the
bino-Vol. 27 • No. 1 • February 1986
Brown- ETHNOBIOLOGICAL NOMENCLATURE
mially labeled classes. But is this hypothesis valid for all biotic forms or only for those indigenous to the environment? For instance, there does not seem to be a one-to-one relationship between low salience and thé use of binomial labels to encode newly introduced food crops like sweet potato and cassava.
When thé sweet potato was introduced to thé Sahu swidden cultivators of northwestern Halmahera (Indonesia), they labeled it a "tuber" after thé yam (lame). Maize is labeled monomially as kaitela, probably because of its Iberian associa-tion. The cassava is classified in Sahu as "maize with a stem"
(kaitela pohong), not as a tuber. However, among thé Bugis of
southern Sulawesi the sweet potato is labeled binomially as a "kind of tuber" (lame kandora) and so is the cassava (lame ayu). It is difficult to say that food crops in général are of low sa-lience, yet thé monomially labeled sweet potato is of little cul-tural value in Sahu, whereas thé binomially labeled cassava is thé staple erop of the Bugis. Hère classification principles other than those which are expressed linguistically seem at issue.
Reply
by CECIL H. BROWN
DeKalb, HL, U.S.A. 27 vm 8S
Much of Berlin's criticism is based on a profound misunder-standing of my position. Berlin writes that I "treat all terminal taxa SL&folk species." What in fact I propose is that terminal generic taxa occurring in biological taxonomies of
hunter-gatherers be relabeled "spécifie" taxa. My reason for this is that
such a change will avoid the awkward characterization of ex-pansion of biological référence as a process involving extending a term for a generic class to a more comprehensive generic class of which the former is a member. In my terminology, a term for a spécifie class will expand to a more comprehensive ge-neric class. Another reason for this change, not mentioned in the paper, is that generic taxa (sensu Berlin), in taxonomies of foragers, tend to bear a one-to-one relationship to scientific species. Indeed, Nancy Turner (personal communication), who has undertaken extensive ethnobotanical research among several foraging groups of the Pacific Northwest, has found my relabeling suggestion sensible and appropriate in view of her discomfort with using the term "generic" in référence to labeled terminal classes which encompass only one species
I do not propose in my paper that all terminal taxa in classificatory Systems of agriculturalists be identified as spécifie classes. In fact, I accept Berlin's framework intact for identify-ing ethnobiological rank affiliation of classes in taxonomies of agrarian peoples. Thus, in my opinion, terminal generics can and do occur in folk taxonomies. I do not, then, "reject" Ber-lin's genera! principles. I only seek to révise them as they apply to biological taxonomies of hunter-gatherers.
Undoubtedly Berlin would object even to a relabeling pro-posai restricted to classes of hunter-gatherer taxonomies be-cause of the importance he and many others before him at-tribute to the psychological (i.e., perceptual) status of generic taxa. In his view, one that, as hè points out, in one form or another has prevailed among biosystematists since Linnaean times, generic classes are "those smallest biological groups readily perceptible to human beings without close study." Al-most certainly Al-most terminal generic taxa sensu Berlin of hunter-gatherer taxonomies are readily perceptible to foragers without close study. Consequently, in Berlin's view, such classes are generic, not spécifie.
